利用Kalman滤波的视频运动目标跟踪

提出一种基于kalman滤波的视频运动目标跟踪算法,首先对视频运动目标进行分割,求出运动目标的形心,再利用视频运动目标的形心所在宏块的运动矢量信息,用kalman滤波对运动目标的形心在下一帧的位置进行预测,从而快速、有效地自动跟踪多个目标对象.实验结果表明,该算法对运动目标的出现和消失,以及非刚性物体的尺度变化和变形,具有较强的鲁棒性.

Tracking Moving Objects in MPEG Videos via Kalman Filtering

In this paper,based on Kalman filtering a video object tracking algorithm is presented.While existing video object tracking is sensitive to the accuracy of object segmentation,the proposed algorithm uses an object central point to complete the object tracking which allows the inaccuracy of object segmentation.Firstly,the authors extract a central point within each segmented object.In tracking step,a motion model is constructed to set system model of Kalman filtering,while the motion vectors of the macroblocks including corresponding central points are also extracted and normalized.Then the position of each central point in the next frame is predicted with Kalman filtering to implement the video object tracking.Since the overall object tracking is carried out via tracking the central point of each object,the proposed algorithm is tolerant to the inaccuracy of object segmentation.Experiments carried out show that the proposed algorithm works well in tracking video objects.